Hybrid Approach for Fake News Detection using CNN and Logistic Regression


  • Abdulazeez Mousa Department of Computer Science, Nawroz University, Iraq
  • Fatih Özyurt Department of Software Engineering, Firat University, Turkey
  • Derya Avcı Department of Computer Technology, Firat University, Turkey




Fake News Detection, Machine Learning, Hybrid Model, Convolutional Neural Networks, Logistic Regression, Explainability


The proliferation of fake news online poses a significant threat to society, eroding trust, manipulating public opinion, and even inciting violence. This paper proposes a novel hybrid approach for fake news detection that combines the feature extraction capabilities of convolutional neural networks (CNNs) with the interpretability and generalizability of Logistic Regression. This synergy aims to address the limitations of both individual methods while achieving improved accuracy and generalizability. Using the Kaggle Fake News Detection Datasets, we rigorously evaluate our model, demonstrating high accuracy in identifying fake news while maintaining interpretability and generalizability. Our research contributes to the field of AI for combating misinformation by developing a more robust and reliable method for fake news detection, paving the way for a more informed and trustworthy information ecosystem.




How to Cite

Mousa, A., Özyurt, F., & Avcı, D. (2023). Hybrid Approach for Fake News Detection using CNN and Logistic Regression. AS-Proceedings, 1(7), 621–627. https://doi.org/10.59287/as-proceedings.755